Where is the test data

In the announcement it said we will use P012, P027, P056 and P065.mha for preliminary submission, however in example code test folder there is nothing, when I run test.sh I get error showing "could not load any files in /input/images/cbct", I wonder do I need to move P012, P027, P056 and P065.mha to /input/images/cbct myself?

Best Qi

Dear qimaqi,

In the announcement it said we will use P012, P027, P056 and P065.mha for preliminary submission, however in example code test folder there is nothing

This is correct; P012, P027, P056, and P065 are the files employed in the test phase. You can find them available under the ground-truth folder (they have a _0000 just to comply with Grand Challenge, but they are the gt labels).

I acknowledge that it was not intuitive how to test the algorithm. For this reason, I have now committed new updates to the repository, including mockup folders and a mockup predictions.json file to make this process easier. Now, you just need to copy your algorithm's output into the respective folder (e.g., the label predicted for F012 should go under ToothFairy2/evaluation/test/F012/output/images/oral-pharyngeal-segmentation).

I hope this makes the whole process easier and more intuitive. Please reach out again if you have any issues.

Best regards,
Luca Lumetti

Moreover, be aware that the test data for the preliminary phase are patients F012, F027, F056, and F065, not P012, P027, P056, and P065.